Jo Feb 16, 2024
When the dimension of data is larger than the number of samples, linear discriminant analysis (LDA) cannot be applied directly to high-dimensional data. This case is called small-sample-size (SSS) or under sampled problem.
To solve this problem, some local learning based various image clustering models are proposed.
Pak Kwang Jun, a lecturer at the Faculty of Applied Mathematics, has proposed a local learning based exponential regularized discriminant clustering model.
In the proposed local exponential regularized discriminant clustering (LERDC) model, local scatter matrices of regularized discriminant model are projected in the exponential domain in order to handle the SSS problem of LDA. In the proposed LERDC model, for each image, the local image matrix is constructed comprising k nearest neighbor images, and the local exponential regularized discriminant model (LERDM) is devised to evaluate the clustering results for the images in the local image matrix.
To verify his method, he compared it with existing state-of-the-art local learning based clustering approaches. The results showed that the proposed LERDC model achieved a comparable clustering performance to that of the near competitor LDMGI model which is based on LDA.
...
Jo Feb 15, 2024
Rare earth neodymium and neodymium-iron alloys are the main materials for production of Nd-Fe-B magnet.
The density, electric conductivity and melting point are the important properties of molten neodymium electrolyte. Knowledge of density is helpful for electrolyte selection with reasonable composition, which contributes to better separation of melts and metallic neodymium.
There have been some reports on the physical properties of neodymium fluoride electrolyte, where alkali fluorides were supposed to replace LiF and the changes of these physical properties with temperature and composition of the electrolyte were considered.
In an attempt to reduce the cost of producing metallic neodymium and neodymium-iron alloys by molten salt electrolysis, Jang Pok Nam, a section head at the Faculty of Metal Engineering, has investigated the melting point, density, electric conductivity and viscosity of ternary (NdF3-LiF-BaF2, NdF3-LiF-CaF2) and quaternary (NdF3-LiF-BaF2-CaF2) neodymium fluorides.
He drew the following conclusions.
The melting point of 90wt% (NdF3-LiF)-10wt% (BaF2-CaF2) melt is the lowest, about 702℃ and the density of molten salts linearly changes with the temperature. In ternary system, the melting points of 85wt%(NdF3-LiF)-15wt%BaF2 and 90wt%(NdF3-LiF)-10wt%CaF2 fall relatively, and in quaternary system, 90wt%(NdF3-LiF)-10wt%(BaF2-CaF2) has the lowest melting point among the considered samples. The relation between temperature and density is linear. Addition of BaF2 and CaF2 increased the viscosities of the considered molten salts. However, 90wt%(NdF3-LiF)-10wt%(BaF2-CaF2) melt has the lowest viscosity. Electrical conductivity of molten salts increases with the temperature and the content of LiF. At the same temperature, quaternary electrolyte has a greater value of electrical conductivity than ternary electrolyte. From the above measurements, it is concluded that 90wt%(NdF3-LiF)-10wt%(BaF2-CaF2) melt can reduce the cost of producing neodymium or neodymium iron alloys.
...
Jo Feb 12, 2024
Ryu Chang Sik, a section head at the Faculty of Distance Education, has developed a large-scale online lecture system in relay mode.
The relay system is an information system capable of providing thousands to tens of thousands of simultaneous subscribers with real-time online education on a nationwide scale in three-tier relay mode through central server, intermediate server, terminal server and terminal on a computer network.
The system aimed to narrow the difference in the level of education between the capital and other provinces and between cities and countries by providing real time or non-real time lectures given by competent lecturers from the capital and other cities to all the schools across the country. Therefore, it has to be designed to transfer real-time lecture data from the capital to provinces, from provinces to counties and from counties to schools, not from the central or provincial server to all terminals.
First, he newly designed and realized a three-stage relaying system via centre-province-county-terminals. Then he solved some technical problems like increasing compression ratio and reducing network load to deal with increase in network traffic per subscriber of three to five times more than the existing video conferencing system, and enhancing the real-time transmission of video and audio data in the relaying mode for perfect performance of online lectures.
Second, he applied load balancing to avoid network overload at provincial, city and county network stations.
Next, as relay servers are installed in all provinces and hundreds of counties across the country, he correspondingly solved the problem of transferring and synchronizing history data between central, provincial and county servers so that a detailed report on attendance in the provinces, cities and counties can be made to the center.
Finally, he made it possible to hold multiple conferences including academic uniformity as well as online lectures.
The proposed system will make a positive contribution to narrowing the differences in education levels between the capital and other provinces by stepping up IT-based education.
...
Jo Feb 10, 2024
Jon Song Won, a researcher at the Faculty of Materials Science and Technology, prepared carbon nanotube reinforced aluminum composite for wire and cables core using carbon nanotubes.
Nowadays, carbon nanotubes, one of carbon nanomaterials, are widely used as reinforcement materials for various composites due to their superior properties.
Carbon nanotube reinforced aluminum composite wires for wire cables produced by new technology can improve mechanical and electrical properties including tensile strength, elongation, bending number and specific electrical resistance more than conventional aluminum wires, and production of wire cables using low purity aluminum is possible.
The preparation process of carbon nanotube reinforced aluminum composite core for wire cables consists of surface treatment of carbon nanotubes, dissolution of aluminum, addition and dispersion of carbon nanotubes, continuous casting and continuous rolling processes.
In order to examine the effect of carbon nanotubes on the texture and properties of carbon nanotube reinforced aluminum composites for wire cables, he performed a tensile test, a repeated bending test, a hardness test and a microstructure analysis.
He drew the following conclusions.
When the content of carbon nanotube is 0.3%, the tensile strength of carbon nanotube reinforced aluminum matrix composites is 250 MPa and the elongation ratio is 5%.
As the carbon nanotube content increases, the particle size of the composite decreases significantly and the shape turns spherical.
His method is more reasonable and economical than the previous methods as there is no need for new equipment for improving the properties of wire and cable cores.
...
Jo Feb 8, 2024
A research team led by Han Jong Su, a section head at the Faculty of Earth Science and Technology, has developed an analysis method of geological structures of irrigation canals and waterway tunnels by magnetic field survey.
Finding out about geological basement is of primary importance for qualitative guarantee of watercourse construction and its safe operation.
Magnetic survey is one of the cheap but speedy and accurate methods.
Magnetic survey based on magnetic susceptibility is considered effective for detecting horizontal and vertical position of boundaries between geological bodies in building a shallow subsurface construction. This method has already contributed to the building of a 100 km watercourse by analyzing the depth and outline of the weathered layers, geological structure and karst.
...
Jo Feb 7, 2024
A research team led by Kim Won Ok, a researcher at the Faculty of Information Science and Technology, has developed a system for detailed survey of forest resources whereby detailed indices such as the number, height, species and volume of individual trees can be obtained from UAV RGB images.
Past surveys of forest resources involved many investigators on the spot, but the system makes surveys easier, scientific and rapid only by UAV.
The system consists of two parts ― one is for obtaining orthomosaic images, digital surface model (DSM) and digital elevation models (DEM) from UAV images and the other is for detecting individual treetops and classifying tree species based on these data in a certain area.
To guarantee the quality of UAV images for survey, it is important to properly determine the photographing parameters of UAV, so through lots of photographing and analysis of UAV images she set photographing interval as 2~2.5s, photographing altitude as 70m from the hillside of the area and air speed as 2.5~3m/s.
This system ensures over 85% of treetop detection rate and over 85% of classification accuracy, and reduces labor for survey down to one sixth.
...
© 2021 KumChaek University of Technology